You may not have heard of Numenta. Perhaps the name Jeff Hawkins is only slightly familiar. That’s Jeff Hawkins, not to be confused with Stephen Hawking.
Hawkins is a true tech pioneer. Founder of Palm, he was instrumental in ushering in the era of the mobile network. But his life long passion with understanding the human brain has led him to his current venture: Numenta.
I wrote a series of article nearly 10 years ago which featured my thoughts on AI, the then fledgling Numenta approach, and Hawkins’ view of how AI ought to proceed. So today, when I read that IBM has now devoted a research center to explore Numenta’s AI algorithms, my curiosity was piqued — and I was, I must admit, quite pleased. Ten years ago I compared Hawkins work with that of Hugo de Garis. I found Hawkins’ thoughts on how AI must necessarily emerge to be compelling, reasonable, and quite probably correct. I read On Intelligence within a few months of publication, and have remained a student of machine learning ever since.
My wordy series, written for Java.net, can be read here in all its broken-link glory:
The Aritsan and the Artilect: part1, part2, part3.
Emergence is the thing. Take the human brain, mimic the abstracted design features in software, and allow the machine to learn. Sensory input becomes patterns becomes knowledge becomes intelligence, provided the abstract characteristics of the human intelligence processing system is accurately captured and duplicated.
But wait … aren’t we courting disaster here? Don’t we have ample cultural artifacts that caution us to not even attempt such feats, lest we surely perish? From Frankenstein to the Terminator to Ex Machina, we have been warned. Stephen Hawking and Elon Musk have signed an open letter with very stern words: AI is our greatest existential threat!
Not so, say others. Many actually involved in AI tell us we should embrace this next inevitable step in evolution. Jeff Hawkins is certainly one of them. So is Tim Oates, a professor of AI, who recently assured us that AI itself is nothing we should fear. As with any technology, read the fine print for acceptable use cases: it’s not the gun, it’s not the bullet, it’s actually the finger pulling the trigger.
Ten years ago I created the quadrant below to depict the Hawkins/de Garis ethical contrast. You should click on the image to read the section from the Artisan and the Artilect entitled The Future of AI. Ten years later, it holds up well … it’s not terrible, and I still believe it to be true.
So contratulations to Numenta! IBM will take it to the next level.
But whether Deep Blue or Deeper Blue, we have nothing to fear. Human competitive general purpose AI is emergent, which means we cannot, strictly speaking, program it. And if it does emerge, we can trust it as a natural and inevitable product of a well-designed universe — sort of like us in the best sense. Will it be conscious? Self-aware? Can it be human-like intelligent otherwise?
Those questions, perhaps, are fodder for another day.
